Skin Segmentation at the Pixel Level Using Fully Convolutional Neural Network

Harsha B K, Indumathi G

Abstract: Modeling skin segmentation is an important activity in applications involving skin detection as a pre-processing element in biometrics, recognition of human activity and many more. There are many challenges faced in the segmentation of human skin pixels like background blend, changing illumination, varying skin races. Therefore, there is a need for proper pre-processing step to remove these redundancies before processing the image for the given application. The pre-processing is achieved by skin segmentation using fully convolutional neural network in this work. The segmentation is evaluated using four datasets and performance parameters are measured.

Keywords: Human skin segmentation, semantic segmentation, Convolutional Neural Network

How to Cite?: Harsha B K, Indumathi G, "Skin Segmentation at the Pixel Level Using Fully Convolutional Neural Network", Volume 11 Issue 2, February 2022, International Journal of Science and Research (IJSR), Pages: 234-237, https://www.ijsr.net/getabstract.php?paperid=SR22202155120, DOI: https://dx.doi.org/10.21275/SR22202155120